《基于神经网络的人体下肢生物动力学模型设计毕业论文(DOC 60页)》由会员分享，可在线阅读，更多相关《基于神经网络的人体下肢生物动力学模型设计毕业论文(DOC 60页)（60页珍藏版）》请在金锄头文库上搜索。
1、基于神经网络的人体下肢生物动力学模型设计作 者 姓 名：葛涛指 导 教 师：王斐单 位 名 称：信息科学与工程学院专 业 名 称：自动化东 北 大 学2021年6月Design of the Lower Limb of the Human Body Biological Dynamics Model Based on neural networkBy GeTaoSupervisor: Associate Professor Wang FeiNortheastern UniversityJune 202150 / 60毕业设计（论文）任务书毕业设计（论文）题目：基于神经网络的人体下肢生物动力学模
2、型设计设计(论文)的基本内容：(1)通过表面肌电信号控制外骨骼运动的背景与发展现状(2)检测下肢表面肌电信号，对信号分析处理提取特征(3)由表面肌电信号与外骨骼输出力矩的关系，运用神经网络建立人体下肢生物动力学模型(4)总结与撰写论文毕业设计（论文）专题部分：题目： 设计或论文专题的基本内容： 学生接受毕业设计（论文）题目日期第周指导教师签字年月日基于神经网络的人体下肢生物动力学模型设计摘要人口老龄化是当今世界各国遍及面临的重大社会问题，下肢外骨骼助力装置的主要功能是扩展人体下肢的运动能力，白叟将是其中的受益者。此外，军事上外骨骼助力装置也可以极大的提高士兵的战斗力，医学上可以辅助病人进行康复
3、训练。将人体划分为若干个具有相同密度和轨则几何形状的刚性肢体环节，各环节之间由球铰连接，并忽略人体组织形变及器官的不对称性，建立人体下肢动力学模型，并计算关节力矩。概况肌电信号是从人体骨骼肌概况记录下来的神经肌肉活动发放的生物信号，能准确反映人的神经、肌肉运动状态。通过人体下肢概况肌电信号，利用神经网络控制外骨骼助力装置的关节力矩，辅助人体下肢运动。本论文主要工作是建立人体下肢动力学模型，运用角动量定理计算人体行走时的关节力矩，通过采集的人体概况肌电信号以及下肢角速度信息，应用神经网络预测关节力矩，实现外骨骼辅助系统力矩输出的智能化。论文首先介绍了外骨骼助力装置的研究意义、国表里发展现状以及本
4、论文的主要研究内容。并介绍了概况肌电信号的形成过程和特点以及人体下肢概况肌电信号与惯性数据采集实验的实验设备，详细记录了实验过程以及实验数据。由于设备信号采集频率分歧，对肌电信号进行了重采样。其次，将人体划分为若干个具有相同密度和轨则几何形状的刚性肢体环节，建立了人体下肢生物动力学模型。对惯性传感器数据进行处理，利用角动量定理，求得关节力矩。再次，介绍了人工神经网络的概念、特点以及在控制范畴的进展，重点介绍了BP神经网络的特点和学习过程，并列出了BP神经网络的改善算法。论文中使用的是LM算法，搭建人工神经网络，经过实验，找到了恰当的隐层节点数。最后，对实验研究进行了总结，网络模型能较好的反映概
5、况肌电信号和关节输出力矩之间的关系。提出了有待进一步解决的问题，并对后来的工作进行了展望。应用概况肌电信号预测关节力矩的准确率还需进一步提高，分外是波峰处的力矩，只要这样，才能实现外骨骼辅助系统的平稳控制。关键词：下肢外骨骼；生物动力学模型；概况肌电信号；关节力矩；神经网络；Design of the Lower Limb of the Human Body Biological Dynamics Model Based on neural networkAbstractPopulation aging is a major social problem the countries in th
6、e world facing. The main function of lower extremity exoskeleton assisting device is to extend the ability of the body movement of the lower limbs and the elderly will be one of the beneficiaries. In addition, the military exoskeletons booster device also can greatly improve the soldiers combat effe
7、ctiveness. In the medical field, it can assist the patient to take rehabilitation The human body is divided into a number of rules that have the same density and geometry of the rigid body link and I ignore human tissues and organs of the asymmetric deformation, and establish lower limb dynamics mod
8、el. Surface EMG was recorded from the surface of human skeletal neuromuscular activity released biological signals. It can accurately reflect the peoples nerves and muscle movement state. Lower limb EMG through the use of neural network control assisting device exoskeleton joint torque, assisted hum
9、an lower limb movement.The thesis mainly studies the biological dynamics model of the lower limb of the human body and the relationship between the lower limb of the human body surface EMG and joint torque. Firstly it introduces the current development of the significance of the research, exoskeleto
10、n device at home and abroad as well as the main research contents of this thesis. It introduces the forming process and the characteristic as well as the lower limb SEMG signal and the inertial data acquisition experimental equipment of surface EMG signal, a detailed record of the experiment process
11、 and the experiment data. Because the signal acquisition frequency of EMG signals, I resample the surface EMG signal. Secondly, the body is divided into a plurality of having the same density and geometry of the rigid body of rules, established the dynamic model of human lower extremity biological.
12、To deal with the inertial sensor data, using the theorem of angular momentum, I calculate the joint torque. And the integration of surface EMG data and joint torque output data laid the foundation for the recognition of surface EMG signal for surface EMG and joint torque. Thirdly, the thesis introdu
13、ces the concept, the characteristics of artificial neural network and the progress in the field of control. It mainly introduces the characteristics, the learning process of BP neural network and the improved algorithm of BP neural network. The paper uses the LM algorithm of artificial neural networ
14、k, set up, after the experiment, found the number of nodes in hidden layer appropriate. Finally, to sum up the experiment, the network model can better reflect the surface EMG and joint relationship between output torque. There are also problems to be solved, and the later work is prospected. The ac
15、curate rate of surface EMG signal control needs to be further improved, when the body of multi-freedom compound action or continuous action, surface EMG signal will be more complex, more difficult to identify the joint torque, higher precision of prediction algorithms need to be further studied.Keywords：Lower extremity exoskeleton; biokinetic models; surface EMG; joint torques; neural network目录毕业设计（论文）任务书I摘要IIAbstractIII第一章 绪论11.1课题研究背景及意义11.2下肢外骨骼装置的国表里研究概况21.3概况肌电信号的国表里研究概况41.4本文主要工作5第二章 下肢运动信息肌电信号采集实验设计72.1概况肌电信号72.1.1概况肌电信号的产生7
